Re: Microtek Phantom 636cx again

Bernd Schroeder (bernd@aquila.muc.de)
Tue, 5 Oct 1999 19:14:20 +0200

Hi,

On Tue, Oct 05, 1999 at 09:30:55AM +0200, Levente NOVAK wrote:
> Hi,
>
> With the help of Bernd, I finally got my Phantom 636cx working. I
> commented out all references to scsi_read_attributes() excepted for
> flatbed mode. Now almost everything is OK, I don't even have that offset
> problem, but instead (nothing is perfect...) I got:
> - the image mirrored (like if it would be seen through a mirror)
> - vertical stripes or bands, especially one much darker. It is like if the
> scanned image would be a copy produced by a dirty copier or a printer
> which has toner problems.
> None of the above problems is present when I scan under Windows. The
> mirror phenomenon is not really annoying as I can flip the images with an
> appropriate program, but the dark stripe is. Does the scanner compensate
> under Windows for the fluorescent tube's light unevenness, or is it
> something else?

If I knew, what the scan software does under Windows, it would probably be
easier to fix problem of a certain type. But normally there is a colour
calibration process at the beginning of each scani which does exactly
such a compensation. According to the
spec this calibration can either be performed by the device or by the
backend. Default is that the device does the calibration, so one could
try the calibration by backend, but this is where the other phantom cx models
stopped working, and it won't work for your particular device anyways.

> Here is the log produced by scanimage for the command "scanimage -d
> microtek2:/dev/sg0 2>backlog.log >image.pgm":
>
> [inquiry]
> 120000006000
> [inquiryresult]
> 060002025b00000020202020202020205363616e6e6572203630304134202020312e33309a0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
> 0: 060002025b000000 2020202020202020 ....[...
> 16: 5363616e6e657220 3630304134202020 Scanner 600A4
> 32: 312e33309a000000 0000000000000000 1.30.... ........
> 48: 0000000000000000 0000000000000000 ........ ........
> 64: 0000000000000000 0000000000000000 ........ ........
> 80: 0000000000000000 0000000000000000 ........ ........
> [scannerattributes]
> 28008200000000002800
> [scannerattributesresults]
> d29100025804b013ec1b3002585d13ec000080ffffff310000003001080200da00e800ed9a000000
>
>
> Scanner attributes from device structure
> ========================================
>
> Scanner ID...
> ~~~~~~~~~~~~~
> Vendor Name: ' '
> Model Name: 'Scanner 600A4 '
> Revision: '1.30'
> Model Code: 0x9a (Phantom 636cx)
> Device Type Code: 0x06 (Scanner),
> Scanner type: Flatbed scanner
> Supported options: Automatic document feeder: No
> Transparency media adapter: No
> Auto paper detecting: No
> Advanced picture system: No
> Stripes: No
> Slides: Yes

This is the reason why the backend aborted, when you first tried it.
Normally the TMA and ADF option can be inquired, if they are supported,
even if there is no TMA or ADF attached to the device. Either this
isn't possible ffor the Slide mode or there is a bug in the backend, so that
it gets confused with stripes and slides. I will check this.

> Scan button: Yes

The Phantom 636cx is probably very similar to the Slimscan C6 which,
according to Microtek's web site, does have a scan button.
Maybe they use the same firmware engine.

> Imaging Capabilities...
> ~~~~~~~~~~~~~~~~~~~~~~~
> Color scanner: Yes
> Number passes: 1 pass
> Resolution: X-max: 600 dpi
> Y-max: 1200 dpi
> Geometry: Geometric width: 5100 pts (8.50'')
> Geometric height: 6960 pts (11.60'')
> Optical resol. : 600
> Modes: Lineart: Yes
> Halftone: No
> Gray: Yes
> Color: Yes
> Depths: Nibble Gray: Yes
> 10-bit-color: No
> 12-bit-color: Yes
> d/l of HT pattern: No
> Builtin HT patt.: 0
> LUT capabilities: None

This attribute normally indicates the size of the gamma and the shading
tables, which can be sent to the scanner. The shading table is used for
the calibration process, so this is the reason, why 'calibration by backend'
will probably not work for you.

Did you try the following: Do one or more scans under Windows, and then
reboot into Linux *without* switching the power of the device off. If
it is a colour calibration problem, it is possible that it is still
calibrated, when you start a scan under Linux.

I don't know, why some devices indicate that they do not accept these tables.
Before I have seen this only once with the device of one V6USL user, but
another device of another user (same model, same FW) had this attribute
set to 4096 words, which is ok.

> Miscellaneous capabilities...
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Data format: Line by line in concatenated sequence,
> without color indicator
> Color Sequence: B - G - R
> Scanning direction: Right to left

This is probably the reason, why the image is mirrored. All other models
I know about, have a scanning direction of "left to right". Currently
the backend ignores this parameter, but it will be relatively easy to fix.

> CCD gap: 0 lines
> CCD pixels: 5100
> Calib wh str loc: -207
> Max calib space: 48
> Number of lens: 1
> Max no of windows: 8
> Sh trnsf func equ: 0
> Buffer type: Ring
> FEPROM: No

[...]

> [readimagestatus]
> 28008300000000000000
> [readimagecmd]
> 2800000000000079ec00
>
> (The last two lines repeated several times)

The last two lines or only the last line ?

Bernd

-- 
Bernd Schroeder 
Email: mailto:bernd@aquila.muc.de
PGP public key available: mailto:pgp@aquila.muc.de | Subject: send key 

--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com